Since none of the proposed emissions limitations for Annex I countries leads to CO2 concentrations that approach stabilization when combined with IS92a or IS92e emissions for non-Annex I countries, and since greenhouse gas concentration stabilization is the ultimate objective of the FCCC, we consider in this Section what additional emissions reductions might be needed to achieve this goal. We do this by comparing the global emissions requirements for stabilization, given and discussed in SAR WGI (Schimel, et al., 1996) and TP3, with emissions under the FR and NL limitation proposals.

In the cases where non-Annex I country emissions follow IS92a or e, concentrations in 2100 range between approximately 575 and 950 ppmv even when the strongest limitation case is considered. The situation is qualitatively different when Annex I country emissions are combined with IS92c emissions for non-Annex I countries (Figure 6). In these cases, by 2100, concentrations under the various emissions limitation proposals increase much more slowly than in 1990 tending towards stabilization at approximately 500 ppmv or less. These results imply that unless population growth, economic growth, technological change and other factors combine in such a way that global emissions mimic the low-emission IS92c scenario, substantial global emissions reductions beyond those defined by the various emissions limitation proposals would be required.

If stabilization is to be achieved, the carbon cycle itself constrains the pathway for global emissions within a relatively restricted range (for any given stabilization target) determined by the concentration pathway (or “profiles") along which stabilization is reached. The differences in emissions between the "S" and "WRE" concentration profiles 16 illustrate this range; further examples are given in Wigley, et al. (1996, Figure 2). When CO2 emissions for different stabilization profiles are compared with those for the various emissions limitation cases, the difference between the carbon cycle emissions constraint and the limitation scenario tells us what additional global emissions reductions are necessary to reach a particular concentration stabilization target. Note that these calculations determine only the additional global emissions reductions that are required. How these additional reductions are apportioned either between non-Annex I and Annex I countries, or across time, depends on political and economic considerations.

16 IPCC has illustrated the effect of concentration pathway on emissions by using two different sets of concentration profiles ("S" and "WRE"). For any given stabilization level, these profiles span a wide range of possibilities. The "S" pathways were defined in Enting, et al. (1994); the "S" stands for Stabilization. The "WRE" profiles were defined in Wigley, Richels and Edmonds (1996) whose initials provide the acronym. Emissions for the "S" series deviate from the central IS92a scenario as early as 1990, whereas emissions for the "WRE" series are constrained to follow IS92a out to 2000 or later depending on the stabilization level.

In Figure 7 global emissions under the FR and NL emissions limitation proposals, with non-Annex I country emissions following the IS92a scenario, are compared with emissions paths that would achieve stabilization at 450, 550 (approximately double the pre-industrial level i.e., 2 x 278 = 556 ppmv) and 650 ppmv. Both the "S" and "WRE" concentration profiles are considered. The emissions results for the stabilization cases are those determined by the Bern model (Siegenthaler and Joos, 1992) and are the same as given in SAR WGI (Schimel, et al., 1996) and TP3. Note that there is currently no agreement about which stabilization level might be appropriate. In SAR WGI (Schimel, et al., 1996) and TP3 additional stabilization levels of 350 ppmv, 750 ppmv and 1000 ppmv are considered. The results presented here represent a middle range of possibilities, easily generalized to other cases.

For stabilization at 450 ppmv, the emissions limitation cases lie between the "S" and "WRE" pathways for the first few decades of the next century, after which they rise increasingly above the emissions for both stabilization cases (Figure 7). Additional, and eventually substantial, reductions in global emissions beyond those given by the limitation scenarios would therefore be required at some time during the early decades of the twentyfirst century if a 450 ppmv stabilization target were to be chosen. Eventual stabilization at concentrations of 550 ppmv and above would permit global emissions to follow any of the proposed limitation pathways at least through the initial decades of the twenty-first century, but substantial reductions below the limitation pathways would still eventually be required. The higher the stabilization target, the longer can the proposed limitation paths be followed and still feasibly attain the target.

20

As a general result, the later global emissions deviate from a particular pathway that does not lead to concentration stabilization (such as IS92a), the larger the subsequent emissions reductions must be in order to achieve stabilization. This principle is clearly demonstrated by a comparison of the "S" and "WRE" emissions paths, noting that the latter follow IS92a initially, while the "S" pathways begin to deviate from IS92a in 1990. In the same way, since none of the emissions limitation proposals when combined with IS92a emissions for nonAnnex I countries approaches stabilization (see Figure 6), the later global emissions deviate from these limitation pathways, the greater the future emissions reduction must be to achieve any given stabilization target. In addition, the longer a particular limitation pathway is followed, the smaller is the cumulative impact on the climate system (i.e., through the reduction in global mean temperature increase or sea level rise) — see, e.g., Wigley, et al. (1996, Figure 3).

These results apply specifically to the case where non-Annex I country emissions follow IS92a. As noted earlier, if nonAnnex I country emissions were to follow IS92c and Annex I country emissions were to follow IS92c or any of the limitation proposals, then stabilization would occur at a level near 500 ppmy with little or no additional intervention. If, however,

Implications of Proposed CO2 Emissions Limitations

non-Annex I country emissions were to follow IS92e, additional global emissions reductions beyond the limitation proposals would be required earlier than in the IS92a case.

These results arise because concentration stabilization requires an eventual reversal of the current upward trend in CO2 emissions, no matter which stabilization target is chosen. For the emissions limitation cases in which non-Annex I country emissions follow IS92c, a reversal occurs early in the twentyfirst century and, by 2100, there is a clear tendency towards concentration stabilization. For emissions limitation cases where non-Annex I countries follow IS92a and IS92e, global emissions rise continuously through the twenty-first century (see Figure 4). Substantial additional emissions reductions are required to reverse the trend. There are economic, social, technological and political constraints on how emissions trends can be reversed, but a consideration of these constraints is beyond the scope of this Paper. For further information, see TP3.

The various emissions limitation proposals also have different implications for climate change, which can be broadly assessed through their effects on global mean temperature and sea level. These effects are considered in the next Section.

To determine the consequences of the various emissions limitation proposals on global mean temperature and global mean sea level we use the models employed in SAR WGI (Kattenberg, et al., 1996). Further details of these models are given in Raper, et al. (1996) and TP2 (Harvey, et al., 1997). These calculations are subject to a number of uncertainties; the primary ones arise from (a) the range of emissions limitation proposals and the assumed emissions for non-Annex I countries; (b) uncertainties in our understanding of the physical processes involved; and (c) choices in how we account for the influences of gases other than CO2:

(a) Range of limitation proposals and assumed emissions for non-Annex 1 countries. To gauge the range of possible effects spanned by the different limitation proposals, we consider only the most extreme case, NL-2%. For nonAnnex I countries we consider three cases: CO2 emissions following the IS92a, c and e scenarios. It is necessary to specify the non-Annex I country emissions because the implications of a reduction in Annex I country emissions depend on the global emissions level. The same reduction has a greater effect on radiative forcing, temperature change and sea level rise when global emissions are small compared with the case when global emissions are large17. In spite of this important effect, over relatively small emissions ranges (such as those spanned by the different limitations proposals at any one point in time), the temperature and sea level responses still vary approximately linearly with global emissions. It is therefore possible to generalize the results presented here by linear interpolation. The results, as noted earlier, also apply to cases where the proposed emissions limitations are expressed in CO2-equivalent terms (see TP3);

(b) Uncertainties in the physical processes. To quantify uncertainties arising from our incomplete understanding of the relevant physical processes, we use a range of model parameter values. This is the procedure used in SAR WGI (Kattenberg, et al., 1996). For global mean temperature, we carry out simulations using three values of the climate sensitivity; viz. equilibrium global mean temperature increases for a CO2 doubling (AT) of 1.5, 2.5 and 4.5°C. All other climate model parameters are as used in SAR WGI, and the simulations use the same slow-down in the thermohaline circulation and the same land/ocean differential climate sensitivity employed in that work. Uncertainties arising from parameters other than the climate sensitivity are relatively small, as demonstrated, for example, in

17 This effect arises mainly because of the non-linear (logarithmic) dependence of radiative forcing on CO2 concentration. For CO2, the concentration reduction caused by a given emissions reduction is actually less for lower global emissions; but this effect is more than offset by the non-linear forcing-concentration relationship.

Wigley and Raper (1993). For sea level we consider low, mid and high ice-melt parameter cases. As in SAR WGI (Warrick, et al., 1996), we combine these with low, mid and high climate sensitivities to better explore the uncertainty range for sea level;

(c) Influence of gases other than CO2. To account for the influences of other gases, we use an idealized "baseline" case for their emissions; specifically, the baseline case used in TP3. Here, the emissions of CH4, N2O and SO2 are kept constant at their 1990 values, and halocarbons follow a scenario consistent with the Copenhagen version of the Montreal Protocol. As in SAR WGI (Kattenberg, et al., 1996), CH4 and N2O emissions are modified from the values given in the IS92 scenarios to ensure a balanced 1990 budget for these gases. This constant 1990 emissions case not only has the advantage of consistency with TP3, but it also avoids complications that might arise because of differences in the emissions of non-greenhouse gases between the IS92 scenarios. Furthermore, by reference to the sensitivity study results given in TP3, it is possible to estimate how the results would be affected by deviations from the constant emissions case for CH4, N2O and SO2 individually.

The global mean temperature and sea level consequences of the NL-2% emissions limitation scenario are shown in Figures 8 to 10. These figures compare "no-limitation" cases, where global CO2 emissions follow the IS92a, c and e scenarios, and "limitation" cases, where non-Annex I country emissions follow IS92a, c and e and Annex I country emissions follow NL-2%. Each figure gives projections for three values of the climate sensitivity (AT2=1.5, 2.5 and 4.5°C), combined with low, mid and high ice-melt estimates for the sea level results.

The influence of the NL-2% emissions limitation scenario in reducing global mean temperature and sea level depends on the emissions case used as the baseline for Annex I emissions (IS92a, c, or e), total global emissions (ie., the emissions assumed for non-Annex I countries), and on the climate sensitivity and ice-melt parameters. Reductions are greater for cases where the Annex I country emissions baseline is higher, since these have larger emissions reductions for any given limitation proposal; and reductions are greater for larger values of the climate sensitivity and/or ice melt. Temperature reductions for the NL-2% case in the year 2100 (low to high climate sensitivity results) are 0.34–0.68°C for IS92a; 0.11–0.23°C for IS92c; and 0.44-0.91°C for IS92e. The corresponding sea level reductions are 4.5-11.5 cm for IS92a; 1.6-4.6 cm for IS92c; and 6.2-15.0 cm for IS92e.

Because of the close empirical relationships between CO2 emissions in a particular year, cumulative CO2 emissions to that year, CO2 concentration and radiative forcing in that year, and

Global temperature change (°C)

the corresponding temperature and sea level values (for any given set of model parameters) in the cases considered here, it is possible to generalize the present results to other cases not considered specifically. The above-mentioned relationships have the characteristic that the temperature and sea level changes to any given year are almost linearly related to the CO2 emissions level in that year. This is shown for the year 2020 in Figure 11. Similar results can be derived for any year, the fit to a linear relationship becomes slightly less good as the date is moved further into the future. To obtain results for 2020 for an emissions limitation case not considered here, the appropriate

Implications of Proposed CO2 Emissions Limitations

2020 emissions value can simply be entered into Figure 11, a climate sensitivity value selected, and an estimate of temperature or sea level change read off using the straight line fitted to the data points. Because all changes are nearly linear in time, it is possible to generalize 2020 results to other years using Figures 8 to 10. These methods of interpolation should, however, be used cautiously; they should only be applied to situations where emissions vary smoothly with time, comparable to the cases considered here.